Noncompetitive N-methyl-d-aspartate receptor antagonists such as phencyclidine and MK-801 are known to impair cognitive function in rodents and humans, and serve as a useful tool to study the cellular basis for pathogenesis of schizophrenia cognitive symptoms. In the present study, we tested in rats the effect of MK-801 on ventral hippocampus (HPC)–medial prefrontal cortex (mPFC) synaptic transmission and the performance in 2 cognitive tasks. We found that single injection of MK-801 (0.1 mg/kg) induced gradual and long-lasting increases of the HPC–mPFC response, which shares the common expression mechanisms with long-term potentiation (LTP). But unlike LTP, its induction required no enhanced or synchronized synaptic inputs, suggesting aberrant characteristics. In parallel, rats injected with MK-801 showed impairments of mPFC-dependent cognitive flexibility and HPC–mPFC pathway-dependent spatial working memory. The effects of MK-801 on HPC–mPFC responses and spatial working memory decayed in parallel within 24 h. Moreover, the therapeutically important subtype 2/3 metabotropic glutamate receptor agonist LY379268, which blocked MK-801-induced potentiation, ameliorated the MK-801-induced impairment of spatial working memory. Our results show a novel form of use-independent long-lasting potentiation in HPC–mPFC pathway induced by MK-801, which is associated with impairment of HPC–mPFC projection-dependent cognitive function.
Cognitive symptoms of schizophrenia, labeled often as impairment of executive function (Elvevag and Goldberg 2000), are largely resistant to the treatment with known antipsychotics. Establishing effective treatment methods for the cognitive symptoms is thus keenly awaited. For this purpose, understanding the cellular and molecular bases of the symptoms would be critical.
Glutamatergic dysfunction exists in many psychiatric disorders such as frontotemporal degeneration and Alzheimer's disease (Hu et al. 2012; Seltman and Matthews 2012), where abnormal glutamatergic neurotransmission may underlie the disease onset. Regarding schizophrenia pathogenesis, a focus also can be made on the “glutamatergic hypothesis,” which derived from the well-known observation that noncompetitive N-methyl-d-aspartate receptor (NMDAR) antagonists phencyclidine (PCP) and ketamine induce cognitive deficits closely resembling schizophrenia cognitive symptoms and exacerbate pre-existing symptoms (Jentsch and Roth 1999). Animal studies equally indicate that the NMDAR antagonists cause deficits of executive cognitive function (Moghaddam and Adams 1998; Stefani et al. 2003; Egerton et al. 2005). In parallel, many schizophrenia susceptibility genes encode factors directly or indirectly involved in the regulation of glutamatergic transmission in the brain (Harrison and Owen 2003).
Evidence indicates that cognitive symptoms in schizophrenia patients involve alterations of the function of prefrontal cortex (PFC) (Weinberger et al. 1994); in particular, an abnormal PFC activity during working memory tasks (Manoach 2003; Driesen et al. 2008). Other studies showed that the NMDAR antagonist MK-801, that possesses a higher selectivity compared with PCP and ketamine, augments mPFC pyramidal neuron discharge possibly through the blockade of NMDARs in local inhibitory interneurons (Homayoun and Moghaddam 2007) and/or inhibitory interneurons in the hippocampus (HPC) (Jodo et al. 2005) that sends monosynaptic projection to the mPFC.
More recently, a hypothesis was put forward that plasticity processes, the capacity of neurons to accommodate use-dependent functional modifications, in the PFC are a cellular target of this mental disease (Goto et al. 2010). Indeed, in vivo human studies indicate that the patients suffering from schizophrenia show impaired glutamatergic plasticity in their brain (Hasan, Nitsche, Herrmann, et al. 2012; Hasan, Nitsche, Rein, et al. 2012). Thus, it is thought that impaired induction of glutamatergic plasticity in critical brain areas such as the PFC might underlie schizophrenia cognitive symptoms by disrupting the activity-dependent adaptations of neuronal connections (Goto et al. 2010).
However, despite the aforementioned observations that support critical roles of NMDARs and glutamatergic transmission in pathogenesis of cognitive abnormalities, it has not been directly tested whether NMDAR antagonists modify synaptic transmission and plasticity in the mPFC. These are important questions since answering to these questions may help us pin down a pathophysiological origin of this devastating mental illness and develop fundamental treatment strategies. We therefore conducted a series of in vivo electrophysiological experiments in which we monitored synaptic responses in the cognitively important HPC–mPFC pathway (Floresco et al. 1997; Marquis et al. 2008; Churchwell and Kesner 2011) under MK-801 challenge. We then tested for the first time the effect of acute MK-801 administration on attentional set-shifting task (ASST) as designed by Birrell and Brown (2000), as well as on delayed spatial alternation (DSA) task (Zahrt et al. 1997), in order to show that MK-801-induced synaptic changes are accompanied by impaired mPFC-dependent cognitive performance. We furthermore sought a clinically relevant method to reverse observed modifications by testing the effect of the subtype 2/3 metabotropic glutamate receptor (mGluR2/3) agonist LY379268.
Materials and Methods
Animals and Surgery
Male Sprague-Dawley rats (Janvier, Le Genest St Isle, France, 350–400 g) were housed in home cage (3 rats per cage). The animal room was maintained on a 12-h light–dark cycle (light was on at 8:00 AM). The rats had ad libitum access to food and water in the home cage. During behavioral experiments, rats were placed on a restricted diet (resulting in 85–90% of their initial body weight) to assure motivation to food reward. The experimental procedures and care of animals were in accordance with the local committee guidelines and the European Communities Directive of 24 November 1986 (86/609/EEC).
Rats were anesthetized by intraperitoneally (i.p.) injected urethane (1.5 g/kg) and placed in a stereotaxic frame (Kopf, CA, USA). Their body temperature was maintained at 37 ± 0.1 °C by a rectal thermometer connected to a homoeothermic blanket control unit (Harvard Apparatus, France). Their skull was exposed, and 2 small holes were made by a dental drill for the insertion of a stimulating and a recording electrode. By the use of manual manipulators (Narishige, Japan), a recording electrode (Teflon-coated tungsten wire, the external diameter 200 µm, A-M Systems, WA, USA) was placed in the midlayer of prelimbic area (3.3 mm anterior to the bregma, 0.8 mm lateral to the midline, 3.5 mm from brain surface), and a stimulating electrode (bipolar Teflon-coated stainless steel wire, the external diameter 200 µm) in the ventral HPC (5.8–6.2 mm posterior to the bregma and 5.3–5.6 mm lateral to the midline, and 4.0–6.0 mm from brain surface). The electrode positions were adjusted to obtain the maximum amplitude of postsynaptic potential (PSP), whose peak appears with an 18.0–24.0 ms delay after stimulus artifact (Laroche et al. 1990). The intensity of stimulation was set to evoke 60% of the maximum response to start experiments.
Response Recording, Plasticity Induction and Drug Injection
The HPC–PFC PSPs were evoked at 0.033 Hz by delivering constant current, monophasic square pulses of 250 µs width (A360 stimulus isolator, WPI, FL, USA) and fed to a differential AC amplifier (model 1700, A-M Systems, WA, USA) with 100 times amplification and the filtration set at 1.0 Hz and 5 kHz. The signals were digitized at 10 kHz through a Digidata 1322A interface (Molecular Devices, CA,USA), and recorded and stored by the use of Elphy data acquisition-analysis program developed by Dr G. Sadoc (Institut Alfred Fessard, CNRS, France) in a PC computer for later analysis.
After a stable baseline response recording for at least 30 min, MK-801 was systemically injected (0.1 mg/kg, i.p.), or in some cases, tetanic stimulation was delivered through the hippocampal electrode (a train of 50 pulses delivered at 250 Hz, repeated 10 times at 0.1 Hz; 2 such episodes in 6-min interval). This dose of MK-801 was chosen because it is subthreshold to visible locomotor disturbance (our routine observation) or causes, if any, very minor stereotypy (Jackson et al. 2004), but is sufficient to impair executive cognitive function (Jackson et al. 2004; Stefani and Moghaddam 2005) independent of the possible subtle locomotor changes (Verma and Moghaddam 1996). Post-MK-801/tetani PSPs were followed for at least 2 h.
Microinjection through a cannula (50 µm diameter), fixed with the recording electrode, was used to apply locally the NMDAR antagonist dl-2-amino-5-phosphonopentanoic acid (AP5). The cannula was connected via a polyethylene tubing to a microinfusion pump (Precidor type 5003, Infors HT) allowing drug infusion while performing electrophysiological recordings. Once a stable signal was obtained, baseline responses were recorded, and artificial cerebrospinal fluid (ACSF; composition in mM: NaCl 124, KCl 2, NaHCO3 26, KH2PO4 1.15, MgCl2 1, CaCl2 2, and d-glucose 11, pH 7.4) or AP5 (200 µM in ACSF) was delivered for 30 min (20 min before and 10 min after the first tetanus or MK-801 injection) at a flow rate of 2 µL/min.
The drugs used in the present study were as follows: (+)-MK-801 maleate (0.1 mg/kg; Tocris); SL327 (10 mg/kg; Tocris); LY379268 (3 mg/kg; gift from Eli Lilly); AP5 (200 µM; Tocris). Drugs that were i.p. injected were dissolved in physiological saline (0.9% NaCl) or DMSO (in the case of SL327) just before the use by a volume of 1 mL/kg. AP5 was dissolved in ACSF just before the local infusion in mPFC.
Attentional Set-Shifting and Delayed Spatial Alternation Tasks
Methods for the ASST and DSA task were adopted from Birrell and Brown (Birrell and Brown 2000) and Wang and Cai (2006), respectively. These tasks requiring the intact PFC were designed to test, respectively, cognitive flexibility and spatial working memory.
In ASST, the testing apparatus was a rectangular plastic arena (inner dimensions, length–width–height: 70–45–60 cm) in which a guillotine door separates the start site. To begin each trial, one rat was placed in the box. At testing, 2 digging bowls were placed. The stimulus bowls consisted of small terracotta pots (internal rim diameter 7 cm; depth 6 cm). Each pot was defined by a pair of cues along 2 stimulus dimensions; that is, the digging medium with which the pot was filled and an odor. To mark each pot with a distinct odor, scented aromatic powder was applied to the inner rim. A different pot was used for each combination of a digging medium and an odor, and only one odor was applied to a given pot. The bait, which was hidden in the “positive” pot and buried with the digging medium, was a 1/3 Chocapic (Nestle®). In all discrimination trials, a small quantity of powdered Chocapic was sprinkled onto the digging medium in the unbaited pot to eliminate the possibility that the rat may locate the bait by smell rather than by learning the discrimination. The procedure of the experiment was as follows:
Habituation (1 week)
In each day, rats were first handled by the experimenter for 10 min in their home cage and then received Chocapic. Later in the same day, each rat was handled alone first and then habituated by 2 to the open-field where they stayed for 30 min. Unbaited pots with sawdust were already present. Then, each rat was placed alone in the open-field with baited pots for 30 min, and the pots were rebaited every 5 min. At the final step of habituation, rats were trained to dig reliably in the pots to obtain a food reward. The bait was covered with an increasing amount of sawdust. When the rats approached rapidly to dig in the pots, they were ready to proceed to the next learning phase.
Learning and Test (1 day)
We injected either vehicle or MK-801 (0.1 mg/kg) 1 h prior to test rats on a series of increasingly difficult discriminations. On each stage, testing continued until the rat reached the criterion of 6 consecutive correct trials, after which testing proceeded to a next stage. First, the rats were subjected to a series of simple discrimination (SD), to a criterion of 6 consecutive correct trials. For these trials, the rats had to learn to associate the food reward with an odor cue (e.g., cinnamon vs. cumin, both pots filled with sawdust). Therefore, in the SD stage, we could confirm that MK801 had no general effects on task learning. Second, they were addressed to the compound discrimination (CD), where the second digging medium was introduced as an irrelevant stimulus. In this stage, only 1 odor was associated with reward, as in the SD stage, but now 2 different digging media were paired randomly with the odors. The third stage was a reversal of this compound discrimination (CDR). In this stage, the same odors and media were used where odor was still the relevant dimension, but the odor-reward contingency was reversed so that the negative odor of the preceding stage became positive, and the positive odor became negative. In the fourth stage, the intradimensional shift (ID) was introduced, where odor was still the relevant dimension while medium was irrelevant, but all stimuli used were novel to rats (i.e., novel odors and novel media). Fifth, a reversal of the intradimensional shift (IDR) was performed and followed by the extradimensional shift (ED). In this stage, using novel sets of stimuli, the relevant dimension was changed so that now the digging medium was the relevant dimension. The final seventh stage was a reversal of the extradimensional shift (EDR) where the medium-reward contingency was reversed. During trials, the correct and incorrect responses were recorded, and rats were allowed to search until they dug in the correct pot to receive reinforcement to facilitate learning. Trials were continued for each test until the criterion level of 6 consecutive correct responses was achieved.
DSA task was used by Zahrt et al. (1997) to test PFC function. We chose this task particularly because the HPC–mPFC pathway is necessary for the performance of this task (Wang and Cai 2006). For the habituation phase, we adopted the protocol of Deacon and Rawlins (2006).
During handling period (1 week), the food quantity for rats was restricted to keep their body weight at 90% of their initial weight. Then, the rats were habituated by pairs before being placed individually in a Y-Maze (length–width–height: 40–15–35 cm) where both goal arms were baited. The rats needed about 5 handling/maze habituation days to be able to run rapidly to each arm end and eat the 1/3 Chocapic in a small dish. After this period, we started the DSA training which consists of a repetition of sessions, where each is a block of 11 trials. The first trial is informative, and it was not included in data analysis. On this informative trial, the rats were rewarded for entering in either arm, but for the 10 following trials, entering in the other arm that was not chosen in the preceding trial was rewarded. Rats were kept in his home cage for 2 min between sessions to avoid mnemonic interferences and task saturation. During the rule training, no delay was imposed between trials, denoted as “0” s delay; all animals were trained at the same time each day in a quiet room. The animals were kept at the end of the side arm for 3 s for reinforcement. In general, 3 sessions were required during 3 consecutive days for the rats to optimally perform the task (the average of correct choice for 3 sessions on the third day was between 90% and 100%). The maze arms were cleaned with 10% alcohol to eliminate odor cues after each trial. After the rule acquisition, delay training was started to determine the maximum delay with which a rat could perform the task at stabilized performance levels at ∼80% correct. The delay in a daily session was constant and was increased by 5 s step if the rat attained a level of ≥80% correct performance in each of 3 preceding successive sessions. When we determined the specific individual maximum delay, we started a baseline procedure which consisted of undergoing 3 daily sessions. In these sessions, 3 variable delay intervals were pseudorandomly assigned over 30 trials: that is, “0” s, half-maximum, and maximum delays. Once rats reached a stabilized alternation score at the criterion of 80% correct and showed at each delay the variation <10% in 3 consecutive days, we tested the rats on the fourth day. We injected either vehicle or MK-801 (0.1 mg/kg) 1 h prior to test sessions. In the second group of rats, we tested whether we could block the effect of MK-801 on the task performance by the mGluR2/3 agonist LY379268 (3 mg/kg), injected 1 h prior to MK-801 injection. We also tested if spatial working memory was still disrupted 24 h after MK-801 injection.
The amplitude of the negative peak relative to the baseline before the beginning of deflection was measured for each response. To express increases/decreases of the response in an individual experiment, we normalized the raw amplitude values by the mean amplitude calculated from the 30-min baseline period just before drug injection/tetanic stimulation. The averaged increase/decrease during the last 15 min of 2-h postepisode recording was taken as the value for plasticity induction (denoted as “2 h after episode”). The percentage values were grouped for each successive 2-min period (i.e., 4 responses) to reduce variability. Analysis of variance (ANOVA) was used to compare postepisode PSP changes and behavioral data between groups. Whenever a significant effect was noted, the data were examined, for electrophysiological experiments, by two-tailed t-test to compare percent changes occurring during the last 15-min period of 2 h recording between groups with P < 0.05 deemed significant (by the use of Statistica 6.1 for Windows). For behavioral experiments, Newman–Keuls post hoc test was used (P < 0.05 as significant; Statistica 6.1 for Windows). All values were expressed as the mean ± standard error of the mean (SEM).
Single Injection of MK-801 Enhances HPC–mPFC Synaptic Responses
After stable recording of baseline responses for at least 30 min from the prelimbic area (referred to as mPFC) upon stimulation to ventral HPC (Gurden et al. 1999; Kamiyama et al. 2011), MK-801 (0.1 mg/kg) or an equivalent amount of vehicle was intraperitoneally (i.p.) injected. This dose of MK-801 was chosen because it is subthreshold to locomotor disturbance (our routine observation) but sufficient to cause impairments of executive cognitive function (Jackson et al. 2004; Stefani and Moghaddam 2005), which is unrelated to, if any, subtle changes in locomotor activity (Verma and Moghaddam 1996). The amplitude of the HPC-evoked mPFC PSP in MK-801-injected rats showed spontaneous and gradual increases (Fig. 1), which reached a plateau in 30–60 min after MK-801 injection (63.2 ± 10.5% increase 2 h after injection, n = 12). Vehicle-treated control rats showed stable PSP throughout the course of the experiment (8.8 ± 6.7% at 2 h, n = 10). ANOVA indicated a highly significant main effect between these 2 groups (F1,20 = 14.167, P < 0.005) and a group × time interaction (F59,1180 = 1.713, P < 0.001) during 2-h postinjection period. The difference of the percent increases in the PSP 2 h after drug/vehicle injection between these 2 groups was also significant (P < 0.0005, two-tailed t-test). Thus, single injection of MK-801 induced spontaneous potentiation of HPC–mPFC responses.
MK-801-Induced Potentiation Shares the Common Expression Mechanism With Tetanus-Induced LTP
In separate naïve rats, we first verified that tetanic stimulation induces long-term potentiation (LTP) in HPC–PFC synapses. Tetani (a train of 50 pulses delivered at 250 Hz, repeated 10 times at 0.1 Hz; 2 such trains applied by a 6-min interval; as used by Gurden et al. 1999 in HPC–mPFC projection) led to a significant LTP of HPC–mPFC responses (67.9 ± 12.0% increase 1 h after and 69.0 ± 14.5% 2 h after tetani, n = 11, Fig. 2A), similar in degree to MK-801-induced potentiation (F1,21 = 1.256, P > 0.2 by ANOVA). We then tested whether MK-801-induced potentiation shares the common expression mechanism with the standard LTP. Thus, in 6 of the 12 MK-801-treated rats, tetanic stimulation was delivered 2 h after MK-801 injection, when the response potentiation had reached a plateau. The responses were followed for further 1 h to evaluate LTP induction. In contrast to naïve rats, in the pre-MK-801-treated rats, the tetani induced no clear LTP, showing only 18.1 ± 13.2% additional increases as measured 2 h after tetani (n = 6, P > 0.05, paired t-test with baseline; Fig. 2B). ANOVA indicated a significant difference between this group and the above control LTP group of naïve rats during the 2-h post-tetani period (F1,15 = 8.160, P < 0.01). Two-tailed t-test also indicated a significant difference in the response increases 2 h after tetani between these 2 groups (P < 0.05). Thus, the prior induction of MK-801-induced potentiation occluded a subsequent LTP by tetanic stimuli.
But this occlusion is undermined since MK-801 is a NMDAR antagonist, which may block LTP regardless of the prior MK-801-induced potentiation (Fig. 4A). To exclude this possibility, we reversed the order of the events. We first induced LTP by tetanic stimuli. Three hours later, when LTP reached a plateau, MK-801 was injected. MK-801-induced potentiation was severely reduced under this condition (17.0 ± 8.4% 2 h after injection, n = 7; Fig. 3A). ANOVA indicated a significant difference between this group and the group in which MK-801 was injected without prior LTP, during 2-h postinjection period (F1,17 = 7.593, P < 0.05). The response change 2 h after MK-801 injection in the prior LTP group (17.0 ± 8.4%) was significantly smaller than the normal MK-801-induced potentiation (63.2 ± 10.5%, P < 0.02, two-tailed t-test). A similar occluding effect by a prior presence of LTP occurred also to a subsequent second induction of LTP by tetanic stimulation applied 3 h after the first LTP, where the degree of the second LTP (20.7 ± 9.4%, n = 5) was statistically indistinguishable from the MK801-induced potentiation after the establishment of LTP by tetanic stimuli (F1,10 = 1.160, P > 0.3 compared with tetani + MK-801 group by ANOVA), further strengthening the possibility that these 2 forms of LTP share the common mechanism.
The above occlusion of MK801-induced potentiation by a prior LTP did not result from a saturated response size caused by the prior LTP since potentiation was still absent when the response size was lowered to the baseline level before MK-801 injection (−19.4 ± 8.1% at 2 h, n = 6; F1,15 = 13.753, P < 0.005; Fig. 3B).
Furthermore, as LTP in HPC–mPFC pathway is NMDAR-dependent (Jay et al. 1995), we tested the effect of local infusion in mPFC the selective and competitive NMDAR antagonist AP5 or the ACSF vehicle (Fig. 4). AP5 infusion prevented the tetanus-induced LTP (9.5 ± 2.6%, n = 5 vs. 62.2 ± 7.6% in ACSF control rats, n = 6; F1,9 = 38.109, P < 0.001; Fig. 4A). In separate rats, local infusion of AP5 blocked MK-801-induced potentiation (11.4 ± 11.7%, n = 5, at 2 h after MK-801 vs. 71.7 ± 9.8%, n = 6, in the ACSF group; F1,9 = 11.511, P < 0.01; Fig. 4B). Thus, both forms of potentiation of synaptic responses in HPC–mPFC pathway are NMDAR-dependent, further suggesting that MK-801-induced potentiation shares the common mechanism with the standard form of LTP.
The above occlusion and local AP5 experiments also suggest that the site of induction of MK801-induced potentiation is the mPFC (see Gurden et al. 2000 for induction locus of HPC–mPFC synapse LTP). The absence of MK801-induced potentiation by the manipulation within the mPFC equally indicates that the induction of MK801-induced potentiation is not a simple response augmentation resulting from a larger number of discharged HPC neurons upon stimulation, for example, due to a local disinhibition in the HPC (Jodo et al. 2005).
Moreover, the above occlusion excludes the possibility that MK-801-induced potentiation is an expressed local synaptic disinhibition in the mPFC through the drug's putative preferential action on local GABAergic interneurons (cf. Jodo et al. 2005; Rotaru et al. 2011), since tetanus-induced LTP is not accompanied by decreased local GABAergic inhibition (Abraham et al. 1987).
MK-801-Induced Potentiation is Independent of Synchronized Synaptic Inputs
The MK-801-induced LTP-like potentiation was unexpected, because LTP is thought to underlie physiological memory function (Malenka and Bear 2004) whereas this dose of MK-801 (0.1 mg/kg) impairs mPFC-dependent behavior (Homayoun et al. 2004; Stefani and Moghaddam 2005). But unlike LTP, the MK-801-induced potentiation required no conditioning stimuli. This indicates that it may be an aberrant form of plasticity with an overly low induction threshold. To strengthen this view, we interrupted 0.033-Hz test stimuli starting at the time of MK-801 injection for 1 h, which is approximately when the MK-801 effect reached a plateau (Fig. 1). Upon a resumption of test stimuli, synaptic responses were potentiated (32.5 ± 3.0% 1 h after stimulus resumption, i.e., 2 h after MK-801, n = 5; Fig. 5) albeit nonsignificantly smaller than the normal MK-801 effect (63.2 ± 10.5%, n = 12, F1,15 = 2.744, P > 0.1). This response increase was significantly larger than changes in vehicle control group that received the identical stimulus interruption (0.7 ± 2.4% 1 h after stimulus resumption, n = 7, F1,10 = 73.034, P < 0.0001). Thus, a significant degree of potentiation was induced by MK-801 without exogenously applied synchronized inputs unlike the standard form of LTP. Consequently, we suggest that MK-801-induced potentiation is an aberrant form of plasticity.
MK-801-Induced Potentiation Requires ERK1/2 Activation
Many forms of synaptic plasticity in the PFC and HPC require postsynaptic activation of extracellular signal-regulated protein kinases 1 and 2 (ERK1/2) (English and Sweatt 1997; Kolomiets et al. 2009; Bai et al. 2014). We therefore tested whether the present MK-801-induced potentiation involves ERK1/2 activation. We injected the specific extracellular signal-regulated kinase activator kinases (MEK) inhibitor SL327 (10 mg/kg, i.p., n = 8), which specifically inhibits ERK1/2, 60 min before MK-801 injection. Control experiments where SL327 was injected 1 h prior to the vehicle followed by response monitoring for 2 more hours verified the stability of responses (2.4 ± 5.3% 1 h after SL327, 2.4 ± 5.3% 2 h after vehicle injection; n = 5, Fig. 6A). Under this condition, MK-801 (0.1 mg/kg, injected 1 h after SL327) induced a severely reduced response increase (14.4 ± 6.3% 2 h after MK-801, n = 8; Fig. 6A), significantly smaller than the normal MK-801-induced potentiation (63.2 ± 10.5%; F1,18 = 9.432, P < 0.01). We verified that i.p. injection itself did not affect MK-801-induced potentiation by injecting the equivalent amount of vehicle followed by MK-801 injection 1 h later (43.1 ± 6.3% 2 h after MK-801, n = 5; F1,11 = 13.113, P < 0.005 compared with SL327 + MK-801 group; Fig. 6A).
However, as ERK1/2 are involved in many important cellular processes, the blockade of MK-801-induced potentiation by a systemic pretreatment with a ERK1/2 signaling inhibitor cannot be applied for clinical use. Thus, we chose to investigate the effect of mGluR2/3 agonist in the following experiments.
MK-801-Induced Potentiation is Blocked by Subtype 2/3 mGluR Agonist
The agonists of mGluR2/3 have clinical importance to treat psychiatric disorders/states including the positive symptoms of schizophrenia (Patil et al. 2007), hyperlocomotion by amphetamine, PCP, and ketamine (Moghaddam and Adams 1998; Pinkerton et al. 2004; Galici et al. 2005), and MK-801-induced PFC neuronal hyperactivity (Homayoun et al. 2005). It was of great interest therefore to test whether MK-801-induced potentiation could be blocked by mGluR2/3 agonists, particularly even when injected systemically. The systemic injection method, although clinically valuable, would not allow one to easily pinpoint the site of the action of drugs. But as mentioned above, the mGluR2/3 agonist appears to act directly on mPFC neurons (Homayoun et al. 2005).
We injected the selective mGluR2/3 agonist LY379268 (3 mg/kg, i.p., n = 5) 1 h before MK-801 injection. LY379268 did not change the baseline responses (8.3 ± 13.3% 1 h after injection; Fig. 6B) compared with the vehicle control (Fig. 1; 9.5 ± 3.9%, F1,13 = 0.479, P > 0.5). Control experiments where LY379268 was injected 1 h prior to vehicle followed by response monitoring for 2 more hours confirmed stable responses (6.0 ± 1.6% 1 h after LY379268, 2.0 ± 2.5% 2 h after vehicle injection, n = 7, Fig. 6B). Under this condition, LY379268 blocked MK-801-induced potentiation (1.6 ± 1.5% 2 h after MK-801, n = 5; Fig. 6B). There was a significant difference between this group and the vehicle + MK-801 injection group (43.1 ± 6.3%, n = 5, Fig. 6A, F1,8 = 18.221, P < 0.004), while there was no significant difference between this LY379268 + MK-801 group and LY379268 + vehicle control group (n = 7, F1,10 = 1.888, P > 0.2; Fig. 6B).
In these LY379268 + MK-801 rats, we delivered tetanic stimuli 3 h after MK-801. LTP was clearly induced (39.3 ± 15.7%, n = 5, Fig. 6B). This LTP was statistically indistinguishable from LTP in the tetanus control group in Figure 2A (F1,14 = 1.033, P > 0.3), suggesting that LY379268 spares the induction of normal LTP. This LTP also did not differ from LTP induced in the LY379268 + vehicle group (48.6 ± 7.6%, F1,10 = 0.574, P > 0.4), confirming that LY379268 injection itself does not affect LTP. Since 4 h had elapsed between LY379268 injection and the tetanus delivery in these LY379268 groups, we verified, in separate animals, LTP induction under LY379268 by delivering the tetani 1 h after LY379268 alone (Fig. 6C). LTP was still inducible (71.6 ± 21.7%, n = 5 vs. 78.2 ± 8.8% in vehicle + tetanus group, n = 7; F1,10 = 0.002, P > 0.9). Thus, LY379268 selectively prevents MK-801-induced potentiation. This result points to the clinically important possibility that systemically administered LY379268 inhibits a spontaneous, use-independent form of plasticity in the HPC–mPFC pathway while preserving a more physiological, use-dependent plasticity induction.
MK-801 Impairs PFC-Dependent Executive Cognitive Function
It is well documented that systemic MK-801 injection causes impairments of mPFC-dependent executive function (Homayoun et al. 2004; Stefani and Moghaddam 2005). We therefore tested whether the present MK-801 treatment also would lead to impairments in mPFC-dependent cognitive function, first by adopting ASST, a well-established rodent version of Wisconsin Card Sorting Test (WCST), which requires the intact mPFC (Birrell and Brown 2000) (see Materials and Methods for details). Specifically, the ASST includes the mPFC-dependent ED set-shifting phase where animals must learn a change of rule. Therefore, this phase is the phase we were particularly interested in order to draw a correlation between MK-801 effects on PFC plasticity and changes in cognition. However, to simultaneously test if MK-801 can affect other discrimination phases, we administered MK-801 (0.1 mg/kg, i.p.) 1 h before the SD, the initial phase of the behavioral test (Fig. 7). Since the whole course of the experiment took <3 h, it was still assured in this procedure that MK-801 had effects at the time of ED test, given that the MK-801-induced potentiation lasted more than 3 h after injection (74.7 ± 14.7% at 3 h, n = 8; 67.8 ± 9.1% at 5 h, n = 4). MK-801- and vehicle-treated rats (n = 8 for each) showed the similar degree of performance up to ID reversal (IDR) phase, demonstrating that MK-801 injection did not disrupt the ability to learn a rule. But MK-801-treated rats showed deficits in ED phase, needing more trials to reach the criterion compared with the vehicle control. ANOVA repeated-measures (stages of discrimination as a within-subject factor and group as a between-subject factors) of trials to criterion revealed a significant interaction (F6,84 = 2496, P < 0.05), and Newman–Keuls post hoc test indicated a significant difference between the groups in the ED phase (P < 0.05, Fig. 7). Thus, MK-801 specifically impaired the mPFC-dependent cognition.
Second, we tested the performance of working memory by adopting DSA task in Y-maze that requires the intact HPC–mPFC projection (Wang and Cai 2006). After 3 days of baseline performance during which the rats showed the correct responses in about 80% of the trials (see Materials and Methods for details), they were injected with MK-801 (n = 8) or vehicle (n = 7) at the fourth day 1 h before the task. First, considering the general performance (combining 3 different delay intervals:“0” s, maximum/2 s, and the maximum delay adjusted in each rat; see Materials and Methods for details), ANOVA analysis indicated a significant difference between vehicle and MK-801 treatments (F1,13 = 13.003, P < 0.004). Indeed, MK-801-treated rats showed significantly less correct responses compared with the vehicle control (P < 0.03, Newman–Keuls post hoc test; Fig. 8A). Furthermore, two-way ANOVA analysis with the delay as repeated measures showed a significant interaction (F2,26 = 7.597, P < 0.03). Specifically, MK-801 injection induced impairment in half-maximum (P < 0.03) and maximum delays (P < 0.03), but not in “0” delay (P > 0.06, Newman–Keuls post hoc test; Fig. 8B, black columns). Thus, MK-801-induced potentiation in HPC–mPFC pathway was accompanied by deficits of cognitive performance that depends on the HPC–mPFC projection.
MK-801-Induced Impairment of Cognitive Function is Ameliorated by Group 2/3 mGluR Agonist
We hypothesized that the mGluR2/3 agonist LY379268 would ameliorate the MK-801-induced, mPFC projection-dependent cognitive function deficits since LY379268 blocked MK-801-induced aberrant potentiation (Fig. 6B). As there is no direct evidence yet that HPC–mPFC projection is involved in the ASST, we chose the DSA task that requires intact HPC–mPFC projection (Wang and Cai 2006) to assess LY379268 effect on cognition.
After 3 days of baseline performance, LY379268 or vehicle was injected at the fourth day 1 h before MK-801/vehicle injection, followed by behavioral test 1 h after the MK-801/vehicle. First, we found a significant effect of treatments for general performance (F1,27 = 5.4, P < 0.03, ANOVA). Indeed, LY379268 significantly improved the task performance in MK-801-injected rats (P < 0.03 between vehicle + MK-801 and LY379268 +MK-801 groups, Newman–Keuls post hoc test, Fig. 8C). Furthermore, two-way ANOVA analysis with the delay as repeated measures revealed a significant interaction (F2,54 = 3.55, P < 0.05). More specifically, LY379268 significantly ameliorated the deficit at the maximum delay (n = 9, Fig. 8D, P < 0.0002 vs. vehicle + MK-801 group, n = 7, and P > 0.05 vs. vehicle + vehicle group, n = 7), while LY379268 did not exert significant effects at half-maximum delay (66.7 ± 3.1 vs. 72.9 ± 2.9%, P > 0.4). This is consistent with the hypothesis that when information is retained for periods longer than 20 s (the maximum delay period in our case), it enters the intermediate/long-term memory storage depending on neuroplastic mechanisms such as LTP (Goldman-Rakic 1996). LY379268 itself did not change the task performance at any delay interval (Fig. 8D, 100.9 ± 0.9, 94.4 ± 2.4, and 93.7 ± 5.7% at 0, half-maximum, and maximum delays, respectively). Collectively, these results indicate that LY379268 blocked at least in part the MK-801-induced deficit of HPC–mPFC projection-dependent cognition.
MK-801-Treated Rats Show Normal Cognitive Performance When MK-801-Induced Potentiation Has Decayed
To further demonstrate a possible correlation between MK-801-induced potentiation and a deficit of HPC–mPFC projection-dependent cognition, we assessed the duration of MK-801-induced potentiation to test whether the cognitive behavior recovers when MK-801-induced potentiation has decayed. For this purpose, first, we attempted to induce MK-801-induced potentiation or tetanus-induced LTP 24 h after MK-801 injection. As we have presented above, we could induce neither LTP by tetanic stimuli 2 h after MK-801-induced potentiation (Fig. 2B) nor MK-801-induced potentiation by MK-801 injection 3 h after tetanus-induced LTP (Fig. 3). In contrast, we successfully induced both forms of potentiation 24 h after MK-801 injection (Fig. 9A,B, see Figure legends for statistics). These results indicate that MK-801-induced potentiation decays within 24 h so that the HPC–PFC synapses regain the capacity for a new episode of potentiation.
Under this condition, in separate rats, we conducted the DSA task 24 h after initially confirming the spatial working memory impairments by MK-801 injection (Fig. 9C). We first observed a significant effect of task between days 4 and 5 (F1,10 = 5.980, P < 0.04) but no group difference at day 5 (F1,9 = 0.025, P > 0.8, Fig. 9C). Thus, the spatial working memory performance in DSA task was normal in rats which received a MK-801 injection 24 h earlier (P < 0.05 between the performance in MK-801 injection day and that 24 h after injection, and P > 0.8 between the MK-801 group, n = 6, and the vehicle group, n = 5, 24 h after MK-801 injection, Newman–Keuls post hoc test; Fig. 9C). When each delay was separately analyzed, we found that the MK-801-induced impairments at the middle and maximum delays were no longer present (P > 0.7 and P > 0.8 between the MK-801 group and the vehicle group 24 h after injection, at middle and maximum delays, respectively, Newman–Keuls post hoc test, Fig. 9D). Thus, at 24 h after the acute MK-801 injection, we were able to induce a new potentiation of the HPC–mPFC responses and observed a restoration of spatial working memory. These results strengthen the possible correlation between MK-801-induced aberrant potentiation and the deficit in the HPC–mPFC projection-dependent cognitive function.
We showed that acute injection of MK-801 induces long-term enhancement of synaptic responses in HPC–mPFC pathway, extending the previous findings that systemic injection of NMDAR antagonists increased the firing rate of PFC pyramidal neurons and c-fos expression in these neurons (Suzuki et al. 2002; Homayoun and Moghaddam 2007; Kargieman et al. 2007). The i.p. injection paradigm under in vivo preparation makes it difficult to precisely localize the site of drug action leading to the response potentiation. But we suggest that it is likely that MK801 acts locally on neurons in the mPFC and/or HPC to affect neurotransmission in this pathway specifically (Suzuki et al. 2002; Jodo et al. 2005; Homayoun and Moghaddam 2007, also see below).
Noncompetitive NMDAR antagonists block active NMDAR, and several studies suggest that GABAergic interneurons are preferentially targeted by these antagonists in the mPFC (Homayoun and Moghaddam 2007; Xi et al. 2009; Del Arco et al. 2011). For a relatively low dose, this mechanism seems plausible since cortical fast-spiking GABAergic interneurons, which seem to play important roles in schizophrenia pathogenesis (Lewis et al. 2005), may have more frequently active NMDAR compared with the pyramidal neurons which are slow-spiking cells. On the other hand, LTP in HPC–mPFC pathway is NMDAR-dependent (Jay et al. 1995) (see also Fig. 4A) where it may be plausible to predict that MK-801 should rather prevent synaptic potentiation. These facts raised the possibility that the present MK-801-induced potentiation is a result of local synaptic disinhibition on mPFC pyramidal neurons rather than a genuine plasticity. But we argue strongly against this possibility. Local synaptic disinhibition may indeed be present after MK-801 injection, but we suggest that MK-801-induced potentiation is an induced plasticity that shares the common expression mechanism with LTP, where a local disinhibition may play a triggering role. First, local infusion of PCP or MK-801 in the mPFC does not increase the firing rate of pyramidal neurons in the mPFC unlike intraventral HPC infusion that augmented HPC pyramidal neuron activity (Suzuki et al. 2002; Jodo et al. 2005). In the HPC, interneurons are indeed more sensitive to NMDAR antagonists than pyramidal neurons (Grunze et al. 1996), which is not the case for the parvalbumin-positive GABAergic interneurons in the PFC, the primary interneurons in this area (Rotaru et al. 2011). Second, MK-801-induced potentiation and tetanus-induced LTP occluded each other (Figs 2B and 3). It is likely that LTP expression is not a result of local synaptic disinhibition, and if anything, LTP-inducing stimuli induce LTP also in synapses on local GABAergic interneurons (Abraham et al. 1987). Thus, if MK-801-induced potentiation is an expressed synaptic disinhibition, the prior LTP, which involves no local disinhibition, should not occlude MK-801-induced potentiation. Third, ERK1/2 signaling inhibitor SL327 blocked MK-801-induced potentiation. This also should not occur if MK-801-induced potentiation is an expressed synaptic disinhibition, since NMDAR blockade on interneurons still occurs under kinase inhibition.
What is then the possible mechanism for MK-801-induced potentiation? Studies demonstrated that systemic injection of MK-801 increases dopamine and glutamate effluxes in the mPFC (Wedzony et al. 1993; Lopez-Gil et al. 2007; Roenker et al. 2012), as in the case for tetanus-induced LTP in HPC–mPFC pathway (Gurden et al. 2000). Glutamatergic and dopaminergic inputs converge on the dendrites of the same pyramidal neurons (Sesack et al. 2003), and as the NMDAR, dopamine in the mPFC is necessary for LTP in HPC–mPFC synapses (Gurden et al. 2000). We thus suggest that a concurrent increase in glutamate and dopamine concentrations in the mPFC, which may be caused by a disinhibitory effect of MK-801 (Roenker et al. 2012) through its preferential action on NMDAR in GABAergic interneurons in, for example, the HPC, may act to induce MK-801-induced potentiation, as in the case for tetanus-induced LTP. Such a local induction of the present aberrant potentiation was also supported by our observation that local infusion of AP5, a competitive NMDAR antagonist that targets all available NMDAR (active and inactive NMDAR), blocked both MK-801- and tetanic stimulation-induced potentiation. As MK-801 only blocks active NMDAR, its action is directly dependent on the biophysical properties of subunits composing the NMDAR. Indeed, MK-801 preferentially acts on NR2A-containing NMDAR (Gielen et al. 2009), and a low NR2A/NR2B ratio in the synapse is postulated to facilitate LTP induction (Yashiro and Philpot 2008). Furthermore, pretreatment with ERK1/2 signaling inhibitor SL327 blocked MK-801-induced potentiation. ERK1/2 are involved in the dopamine- and glutamate-dependent plasticity in the mPFC (Kolomiets et al. 2009; Bai et al. 2014), suggesting that ERK1/2 serve as a common downstream pathway for plasticity induction. It would be important to test in the future the effect of local blockade of dopamine signaling in the mPFC, as well as local decrease of glutamatergic neurotransmission, on MK801-induced potentiation by using, for example, an mGluR2/3 agonist.
As the HPC–mPFC pathway critically supports executive function (Floresco et al. 1997; Churchwell and Kesner 2011), we tested the effect of MK-801 on this cognitive function. In humans, cognitive flexibility can be measured by the WCST with which schizophrenia patients exhibit lower performances (Everett et al. 2001). We tested this ability in rats using the ASST (Birrell and Brown 2000) in which MK-801 specifically altered the ED shift, consistent with the previous findings obtained with PCP (Egerton et al. 2005) or ketamine (Kos et al. 2011). As MK-801-treated rats could normally learn the initial task rule (SD stage) and the other stages except the mPFC-dependant ED, the observed performance alterations is likely to be due to plasticity modification in the mPFC and not caused by possible secondary impacts of plasticity on the excitability or activity in other regions. These results indicate that, as seen in schizophrenia patients, MK-801-treated rats are unable to modify a previously adopted rule and/or to inhibit a learned behavior (Pantelis et al. 1999; Tyson et al. 2004; Ceaser et al. 2008).
The involvement of the aberrant HPC–mPFC transmission in the deficits of executive function was strengthened by our second observation that MK-801-injected rats were impaired in the DSA, a task that tests spatial working memory which schizophrenia patients also show deficits (Driesen et al. 2008) and that requires the intact HPC–mPFC projection in rats (Wang and Cai 2006). Working memory may be stored by a maintained firing of a subset of neurons in PFC network (Funahashi et al. 1989) probably through processes involving recurrent microcircuits (Amit and Brunel 1997; Camperi and Wang 1998; Durstewitz et al. 2000; Constantinidis and Wang 2004) and a short-term creation of cell assemblies (Fujisawa et al. 2008) which forms a functional connectivity. Considering this mechanistic aspect, we suggest that MK-801-induced potentiation disrupts working memory by altering a successful creation of such memory fields based on the information conveyed by both HPC inputs and increased dopamine levels in the mPFC (Zahrt et al. 1997; Vijayraghavan et al. 2007), perhaps mechanistically involving decreases of signal-to-noise ratio (Jackson et al. 2004). Finally, since the DSA is easily separated into the initial task-learning phase and the later, learned-task performance phase, it would be interesting in the future to test the effect of MK801 on the initial learning phase. This would isolate the effect of MK801 on the behavior relying on plasticity induction in critical neurons, and help us further rule out the possibility that the current MK801 effects on behavior resulted from the secondary impacts of plasticity on the excitability or activity in other brain regions.
Intriguingly, we found that MK-801 effect on the HPC–mPFC plasticity decayed within 24 h (Fig. 9A,B), because a successive potentiation of responses by the second MK-801 or a tetanic stimulation could be obtained 24 h, but not 2–3 h, after the first MK-801 (Figs 2B and 3). Similarly, at 24 h after MK-801 treatment, spatial working memory performance was completely restored (Fig. 9C,D), showing a correlation between MK-801-induced cognitive deficits and the aberrant plasticity. Nevertheless, we still do not rule out the possibility that increased transmitter efflux in the mPFC per se, rather than the aberrant plasticity, is the cause of cognitive deficits. Also, abnormalities in other pathways, such as intrahippocampal CA3-CA1 transmission, may be equally involved in the cognitive deficits. To confirm the causal relation between the aberrant plasticity in PFC neurons and behavioral abnormalities, it would be necessary to carry out further investigations where, for example, blockade of MK-801-induced potentiation by intra-mPFC infusion of drugs, such as dopamine receptor antagonists, is correlated with the absence of cognitive deficits.
MK-801-induced potentiation was blocked by mGluR2/3 agonist LY379268. Provided that NMDAR antagonists increase glutamate release in the mPFC (Wedzony et al. 1993; Lopez-Gil et al. 2007) and that mGluR2/3 agonists block these increases (Moghaddam and Adams 1998; Lorrain et al. 2003), we suggest that this latter inhibitory action underlies the blocking effect of LY379268 on MK-801-induced potentiation. But this blocking effect is specific to plasticity induction since we detected no decreases in the baseline responses by LY379268, similar to the case seen in neural activity (Homayoun et al. 2005).
It was intriguing that LY379268 blocked MK-801-induced potentiation but spared tetanus-induced LTP, a finding with possible therapeutic importance. We discussed earlier that the MK-801-induced potentiation may depend on increased glutamate and dopamine levels in the mPFC. In this respect, it is noteworthy to mention that systemic injection of LY379268 or LY354740 prevents the NMDAR antagonists-induced increases in glutamate level in the mPFC but not dopamine (Moghaddam and Adams 1998; Lorrain et al. 2003). Therefore, blockade of glutamate enhancement in the mPFC may be sufficient to prevent MK-801-induced potentiation but preserve the capacity for dopamine increases in the mPFC. Combined with this capacity, the use-dependent LTP can be still induced in the presence of LY379268 since tetanic stimuli to the projection fibers might robustly increase glutamate level in the mPFC to overcome the inhibitory action of LY379268 on glutamate increase.
We further found that the blockade of MK-801-induced potentiation by LY379268 was accompanied by an improved performance in the DSA task. Similar beneficial effects of mGluR2/3 agonists have been shown on NMDAR blockade-induced hyperlocomotion and working memory deficits (Moghaddam and Adams 1998; Cartmell et al. 1999). In the present case, our detailed analyses detected a specific and clear recovery of performance in LY379268 + MK-801-treated rats only at the maximal delay. It remains unknown in terms of the memory mechanism how the maximum delay-associated memory appears without initial shorter term memory, but a similar, long delay-specific improvement was seen with LY354740 for PCP-induced deficit of spatial working memory (Moghaddam and Adams 1998).
In summary, we showed for the first time that MK-801 injection induces potentiation of HPC–mPFC responses in rats that shares the common mechanism with tetanus-induced LTP. This aberrant plasticity was accompanied by impairments of mPFC-dependent executive functions. Future studies are required to better clarify the underlying mechanisms of this plasticity as well as to confirm the main finding under freely moving condition, which we believe will help understand the cellular mechanisms of schizophrenia cognitive symptoms. Indeed, pharmacological tools or transcranial magnetic stimulation that would reverse aberrant plasticity should provide a potential clinical significance for the treatment of schizophrenia.
This work was supported by the INSERM, CNRS, UPMC and the French Ministry of Higher Education and Research. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
Conflict of Interest: None declared.